China issues guidance on AI-generated voices and likenesses
Guidance from China’s highest court addresses digital reproductions without consent and offers a starting point for examining identity protection in Brazil.

AI-generated illustrative image. Fictional audio production scene.
A person may recognise their own voice in a recording of words they never spoke, or find their face in a video they never appeared in. These are among the issues addressed in guidance on artificial intelligence released by China’s Supreme People’s Court on 7 September 2026.
The document guides Chinese courts in disputes involving the technology. Section 4 addresses digital likenesses and synthetic voices that make an individual identifiable. Subject to statutory exceptions, it provides protection against creating and using identifiable digital representations without consent, and against using someone’s voice without consent as training material to generate a recognisable imitation.
More than a convincing imitation
The guidance also addresses the use of these representations to attribute false statements or conduct to someone and damage their reputation.
The connection between a digital reproduction and the recognisable individual matters. A familiar voice or face can make a message appear credible to its recipient. A fabricated recording can exploit that trust.
How Brazilian law protects voice and likeness
In Brazil, these questions engage personality rights, a category of civil-law protections concerning the individual. Article 12 of the Brazilian Civil Code allows a person to seek an end to a threatened or ongoing violation and claim damages. Articles 20 and 21 address protection of likeness, the transmission of a person’s words and private life. These provisions provide relevant grounds for examining AI uses that affect personal identity.